New methods for ultra‐high throughput proteomics for biomarker discovery

Abstract

Abstract

Background: Plasma biomarkers are important for detecting diseases and predicting outcomes. However, to ensure reliable biomarker discovery, analyses should be conducted in large sample cohorts. Current plasma-based MS proteomics methods either focus on high-throughput analysis or proteome coverage but approaches that achieve both typically rely on costly workflows. Thus, there is a critical need for plasma proteomics methods that balance coverage, throughput, and data quality to enable more reliable and cost-effective biomarker discovery.

Hypothesis: An optimised clinical plasma proteomics method will generate high-quality data over shorter timescales, enabling broad-scale biomarker discovery.

Aims: (1) To assess the established proteomics workflow, (2) to shorten the LC gradient method while maintaining data quality, (3) to evaluate different DIA software tools, (4) to improve the plasma sample preparation workflow, and (5) to validate the developed method for biomarker discovery.

Methods: This study used a DIA-based plasma proteomics workflow developed on the Zeno TOF 7600 MS. A spike-in strategy, using E. coli protein added to human plasma, was used as a standard to support method development and performance assessment. This workflow was evaluated across different experimental variables, including MS instrumentation and dataprocessing software. Data were processed using DIA-NN with a library-based analysis approach. The optimised workflow was subsequently applied to a cohort of 206 plasma samples collected at 10 time points from 22 living kidney donors following kidney removal for transplantation to assess its applicability to plasma biomarker discovery.

Results: An optimised, non-depletion plasma DIA proteomics workflow was developed using the Zeno TOF 7600 MS. The optimised workflow employed a rapid 21-min LC gradient and the DIA-NN software tool for data analysis, enabling robust, high-throughput analysis of neat plasma samples. Compared with the previously established Triple TOF 6600 workflow, the optimised method demonstrated improved quantitative performance, including reduced missingness (5.8%), lower CVs, with 83.2% of proteins reproducibly quantified at CVs <30%, and enhanced quantitative accuracy through effective detection of true positive changes in protein abundance, while maintaining reproducible proteome coverage. Application of this workflow to plasma samples from living kidney donors revealed biologically significant changes in protein abundance, demonstrating its suitability for clinical plasma biomarker discovery.

Conclusion: Using a standard sample enables optimisation of LC-MS plasma biomarker discovery methods. These methods support the identification of cystatin C, PEDF, CLEC3B, afamin, gelsolin, and complement components as candidate biomarkers of the adaptation process or renal function in living kidney donors.